> > > You can use the kernel profiling to find out. Also remember to compare
> > > non SMP kernels with non SMP kernels
> >
> > Hm. How can dcache be slower and better at the same time? :)
>
> If you look at the numbers the machine spent more clock time doing less
> system time work. So if the dcache is slowing it then its because its causing
> paging. Someone else suggested its the writeback of a deleted inode that
> may be involved
Nope, no paging took place. As I said, lots of free RAM.
Speaking of writing deleted inodes, why wouldn't 2.0.33 write them to
disk, in the first place?
I didn't reboot that machine yet, but if you're correct, I should see
loads of "inode with zero dtime" or similar messages, while doing
fsck? Am I right?
>
> > > Can you see if 2.1.115ac1 is any better on this benchmark on your box
> > I'll check, as soon as I sit physically at my machine (tomorrow).
>
> I'd expect similar slower wallclock but even lower system time
>
Why can't wallclock be lower?
I mean, I didn't check -ac1 yet (waiting for Monday), but I don't see
a reason why shouldn't 2.1 be faster than 2.0. Dcache is definitely
more elaborate in 2.1 so we can only expect improvements, right?
Reagrds,
-- Posted by Zlatko Calusic E-mail: <Zlatko.Calusic@CARNet.hr> --------------------------------------------------------------------- I hate when they fix a bug that I use.- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html